How will robots get to work?
Our community has, for a long-time, been thinking about autonomous vehicles as the future of transport, and in some ways, they still are. But humanoid robots are also coming: see this article in USA Today, obviously there is still a lot of vapourware, but also some impressive technology. Humanoid robots have potential applications related to transport. They have the additional advantage of fitting into environments designed for people.
What makes a robot “humanoid”?
There are many definitions for robots. There is in fact a whole highly entertaining podcast built around this theme: Robot or Not, which examines this question, and word definitions in general. In brief, a robot perceives the physical world, decides, and acts. A humanoid robot does that with legs, arms, and sensors. It walks, climbs stairs, turns knobs, opens doors, and carries things. Most notably, it can operate in spaces designed for people, because much of the human world, especially the insides of buildings and vehicles, wasn’t built for drones or wheels.
This isn’t about some far-off science fiction future, humanoid robots are dropping rapidly in price, and it’s time we start thinking about their implications.
They’re not optimal in some idealised sense. We would not design most tasks to be automated with humanoid robots if humans had not come before; but they are compatible with everything we’ve already built.
Why now?
Robots aren’t new, they have been staples of science fiction (and science fiction comedy) for more than a century. When I worked at Hayes Microcomputer Products in the mid 1980s, there was a giant robotic arm on the factory floor. Humanoid robots doing useful work has long been a dream.
Humanoids are now finally cheap enough, and probably capable enough, to do useful work in the real world. This is because:
Motors and joints are lighter and more reliable.
Battery life is up.
Perception is good enough to avoid bumping into walls.
AI has reached the point where basic movement and decisions can happen without full scripting.
Prices are dropping fast. Entry-level models now cost under $6,000, the NEO looks to be $20k US.
They don’t necessarily replace a person. They can however replace multiple specific tasks: lifting, fetching, walking, waiting. Chain enough tasks together, and it can do what many workers are paid to do.
Some Things
Here are some transport-ish use cases, you might have others:
In the home
Accept packages and bring them inside.
Assist or carry someone up the stairs who can’t climb them.
Collect health samples.
At school
Shuttle materials between rooms and floors.
Support students with limited mobility.
Escort young children between classrooms.
In logistics
Carry goods up stairs in buildings with no lift.
Place items directly on balconies in locked-down buildings. (See the Low-Elevation Economy)
Rebalance scooters or bikes across multi-level sites. (Maybe even robots riding bicycles)
Restock vending machines in awkward spaces.
In transport
Drive unmodified vehicles inside depots or even on public roads.
Repark ferries and buses after service hours.
Guide tourists, carry bags, or translate instructions.
In emergencies and public service
Evacuate non-ambulatory people from buildings.
Patrol for fare compliance or safety.
Help people navigate border checks or quarantine.
Pick up trash after street markets close.
They fill in where humans are scarce or expensive, where risks are high, where stairs break other kinds of automation, or where kerbs aren’t accessible.
What actually changes?
Following traditional travel behaviour analysis, we presume that Robots can:
Replace some trips entirely (e.g. school errands, stair-assist).
Enable trips that didn’t happen before (e.g. a robot bodyguard (no Asimovian 3 laws for you, clearly they can hurt you) securing safe walking at night enables trips that would have been substituted for by staying at home).
Shift the timing of trips (e.g. off-peak delivery).
Expand destinations, letting services reach places they couldn’t (e.g. walk-ups, flood zones).
Robots remove friction, but when does that removal justify the cost? That depends on the operational costs of the robot (presumably small), and their capital cost (relatively large), as well as their lifespan, TBD. But I would not complain if someone got me a robot, for research of course.